Overview

1. Overview • Last Lecture • Packet switching in Wide Area Networks • Source: chapter 10 • This Lecture • Routing in WAN • Source: chapter 10 • Next Lecture • X.25 • Source: chapter 10

2. Routing • Complex, crucial aspect of packet switched networks • Characteristics required • Correctness • Simplicity • Robustness • Stability • Fairness • Optimality • Efficiency • Costs is the major factor in routing • Length of path • Number of hops • Transport time • In-route delays

3. Routing decision • Decision time • When forwarding each packet • When set up a virtual circuit • Decision place • Distributed: made by each node/router • Centralized: made by a central location • Source: made by the sender • Routing decisions usually based on knowledge of network (not always) • Distributed routing • Nodes use local knowledge • May collect info from adjacent nodes • May collect info from all nodes on a potential route • Central routing • A central station collects info from all nodes • Update timing • When is network info held by nodes updated • Fixed - never updated • Adaptive - regular updates • Routing strategies • Fixed, flooding, random, adaptive

4. Fixed routing • Single permanent route for each source to destination pair • Determine routes using a least cost algorithm • Route fixed, at least until a change in network topology

5. Flooding • Scheme • No network info required • Packet sent by node to every neighbor • Incoming packets retransmitted on every link except incoming link • Eventually a number of copies will arrive at destination

6. Flooding • Optimizations • Each packet is uniquely numbered so duplicates can be discarded • Nodes can remember packets already forwarded to keep network load in bounds • Can include a hop count in packets • Properties • All possible routes are tried • Very robust • At least one packet will have taken minimum hop count route • Can be used to set up virtual circuit • All nodes are visited • Useful to distribute information (e.g. routing) • Random routing • Node selects one outgoing path for retransmission of incoming packet • Selection can be random or round robin • Can select outgoing path based on probability calculation • No network info needed • Route is typically not least cost nor minimum hop

7. Adaptive routing • Used by almost all packet switching networks • Routing decisions change as conditions on the network change • Failure • Congestion • Pros • Improved performance • Aid congestion control • Cons • Requires info about network • Decisions more complex, may not realize theoretical benefits • Tradeoff between quality of network info and overhead • Reacting too quickly can cause oscillation • Too slowly to be relevant • Classification based on info sources • Local (isolated) • Adjacent nodes • All nodes

8. Isolated adaptive routing • Route to outgoing link with shortest queue • Can include bias for each destination • Q+B is used to decide the route • Rarely used - do not make use of easily available info

9. Adaptive routing - adjacent nodes • Distance Vector Routing • Developed by Bellman-Ford, also called Bellman-Ford algorithm or backward search algorithm • Used in ARPANET until 1979 • Belongs to distributed adaptive algorithm • Scheme • Initially each router uses ECHO packets to learn the distance to each of its neighbours. • Periodically each router sends each neighbour a list of estimated costs to each destination • Each router updates its tables with the best cost to each destination according to the information received from its neighbours • Problems • Delay of link information • Count-to-infinity problem (Shay’ book,pages 464-465)

10. Adaptive routing - all nodes • Link State Routing • Developed to replace Distance Vector in ARPANET • Variants of Link State routing is now widely used • Similar to Distance Vector routing • It differs from Distance Vector in • what information it exchanges and how to store and process it. • Link information is more timely propagated • Scheme • A node gathers information on the status of each link to each neighbour, e.g. bit rate, delay time reliability, number of queued packets • The node builds a link state packet for each link, containing the above information • A node receiving link state packets forwards them to all its neighbours except the one from which it received the packet • As link state packets are exchanged among nodes, each node eventually learns about the network topology and the cost and status of links between network nodes • Consequently each node can execute a cheapest route algorithm such as Dijkstra’s algorithm to determine its routing table

11. Routing in ARPANET • First Generation • 1969 • Distributed adaptive • Estimated delay as performance criterion • Bellman-Ford algorithm (appendix 10a) • Node exchanges delay vector with neighbors • Update routing table based on incoming info • Doesn't consider line speed, just queue length • Queue length not a good measurement of delay • Responds slowly to congestion • Second Generation • 1979, link state routing • Uses delay as performance criterion • Delay measured directly • Uses Dijkstra’s algorithm (appendix 10a) • Good under light and medium loads • Under heavy loads, little correlation between reported delays and those experienced • Third Generation • 1987 • Link cost calculations changed • Measure average delay over last 10 seconds • Normalize based on current value and previous results

12. Hierarchical Routing • Previous algorithms require to keep routing information of each node • There may be too many nodes for each to have complete routing tables. • Consider a postal address: country, city, street, number • Scheme used by Internet • Nodes are divided into groups called domains, which can be considered to be a separate and independent network • Routes between two nodes in a common domain are determined using the domain's or network's protocols • Each domain has one or more specially designated nodes called routers which determine routes between domains. • Domains may consist of subdomains each with its own router

13. Routing for Mobile Hosts • Problem: to route a packet to a mobile host • Two kinds of users • Migratory users: stationary users who move from one fixed site to another from time to time but use the network only when they are physically connected to it • Roaming users: compute on the run and want to maintain their connections as they move around. • First step: how to find the mobile host? • The world is divided into small areas • Each area has a home agent, which keeps track of users whose home is in the area • Each area has one or more foreign agents, which keep track of all mobile users visiting the area • When a new user enters an area, his computer must register itself with the foreign agent there, so that the foreign agent can inform its home agent where it is

14. Routing for Mobile Hosts (cont.) • Second step: how to send a packet to a mobile host? • When a packet is sent to a mobile host, it is routed to its home agent in the first place • Because the home agent knows the address of the host’s foreign agent, it forwards the packet to the foreign agent which then passes the packet to the mobile host • Then the host’s home agent tells the sender to henceforth send packets to the host’s foreign agent, instead of sending to the home agent • Then subsequent packets can now be routed directly to the mobile host via the foreign agent, by passing the home location entirely

15. Summary • Issues in routing • Cost • Decision time and place • Routing classifications • Distributed • Central • Fixed • Adaptive • Routing algorithms • Fixed • Flooding • Random • Adaptive • Isolated • Distance Vector Routing • Link State Routing • Hierarchical • Mobile